What is run-experiments?

A procedure for running LiDAR experiments unattended for a set period. LiDAR is a sensing method that measures distance with laser light; the procedure chooses questions, runs replay tests on available captures, and records reports and proposed changes.

Use it to select and run product experiments for a chosen duration or area, compare results, make passing changes the default, and produce reports, pull requests, and summaries.

It keeps experiments reproducible by writing the rule before testing and leaves time to document results. It also prevents long-running jobs from blocking the main session.

How banshee-data/velocity.report configures Claude Code

Reuse

Borrowing it

Nothing to install: this file belongs to banshee-data/velocity.report. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/banshee-data/velocity.report/main/.claude/skills/run-experiments/SKILL.md

Clone the repo

git clone --depth 1 https://github.com/banshee-data/velocity.report

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/banshee-data/velocity.report/run-experiments/github.svg)](https://agentmods.dev/skills/banshee-data/velocity.report/run-experiments)

Your own site

<a href="https://agentmods.dev/skills/banshee-data/velocity.report/run-experiments"><img src="https://agentmods.dev/badge/skills/banshee-data/velocity.report/run-experiments/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/banshee-data/velocity.report/run-experiments"><img src="https://agentmods.dev/badge/skills/banshee-data/velocity.report/run-experiments.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 90 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 3,615 The whole file, excluding the scripts and references it only reads on demand.

Security scan C 2 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00090	$0.03615
Opus 5.5	$0.00036	$0.01446
Sonnet 5.5	$0.00018	$0.00723
Haiku 4.5	$0.00009	$0.00362

Measured today against content hash f2593b439e58, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade C, and why

run-experiments scanned grade C with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Recursive force deletehighDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

- Never `--no-verify`, `rm -rf`, `git reset --hard` or a bare `git stash`. Commit with explicit

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

(`curl -sI https://huggingface.co | head -1`); if it does not, run kirk0 only and say so. A whole

.claude/skills/run-experiments/SKILL.md · 216 lines

How it starts

The opening of the file, as written. The whole thing — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Skill: run-experiments

Run experiments unattended until the time is up: choose the questions that move the product most, answer each under a rule written before its replays, make what passes the default, and leave a report, a PR and a summary for each. The method is the one the box rotation and box alignment runs used (geom0005 to geom0010, l5id0001 to l5id0005).

Usage

/run-experiments                 # 12 hours
/run-experiments 5h              # a set duration: h and m, e.g. 90m or 2h30m
/run-experiments 6h l5id         # only this area's questions, or named slugs (l5id0006 geom0011)

1. The clock

Record the start (date -u) and the deadline. Stop starting new arms once the time left is less than the next arm's estimate plus the time to write it up: keep the last hour, or a sixth of a run shorter than six hours, for reports, PRs and the summary. Check the clock between steps, not only at the end. Never wait with a foreground sleep: run long jobs in the background and wait with an until-loop or the Monitor tool, and keep checking disk and the clock while they run.

2. Where the run is

Find the host before choosing anything: it decides the captures, the scratch space and where raw output may go.

hostname; uname -sm; echo "${CLAUDE_CODE_REMOTE:-}"; df -h . /tmp
ls /Volumes/lidar/lidar/pcaps /Volumes/banshee-captures/lidar /mnt/captures/lidar 2>/dev/null | head

Local (the Mac, or any machine with captures attached).

Captures: any PCAP present. LIDAR_PCAP_DIR from local.mk (on the Mac /Volumes/lidar/lidar/pcaps: S2 segments under s2/, whole placements under sf-street-speeds/raw/lidar/), the NAS at /Volumes/banshee-captures/lidar, and kirk0.
Scratch: ~/vr-scratch/<run>/ on the internal disk.
Raw output: archived to /Volumes/banshee-results/velocity-campaign/<run>/, the results share on arrow.local. LiDAR data goes on no other volume.

arrow-worker (the VM bansheeworker on the TrueNAS host arrow).

Read the full file on GitHub · 216 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

today Changed · +25 lines f2593b439e58
14d ago Changed · +2 lines a4cec98727c7
19d ago First seen · 189 lines · 90 tokens per session scan C 3938c89472b6

Subscribe to this mod's changes

run-experiments is a skill published in the GitHub repository banshee-data/velocity.report (11 stars, last pushed today), licensed Apache-2.0. It adds 90 tokens to every session and 3,615 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-18.

Research Scientific lidar radar road-safety street-safety traffic-analysis urban-analytics

run-experiments next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
run-experimentsbanshee-data/velocity.report	Skill	11	today	Apache-2.0	C	90	Claude Code
whitebox-toolsznlgis/opengis-skills	Skill	76	10d ago	MIT	A	119	Claude Code, Codex
pdalznlgis/opengis-skills	Skill	76	10d ago	MIT	B	90	Claude Code, Codex
instrument-data-to-allotropeanthropics/knowledge-work-plugins	Skill	35k	2d ago	Apache-2.0	A	123	Claude Code
exploratory-data-analysisK-Dense-AI/scientific-agent-skills	Skill	56k	4d ago	MIT	A	83	Claude Code
phylogeneticsK-Dense-AI/scientific-agent-skills	Skill	56k	4d ago	MIT	A	71	Claude Code, Codex
research-engineerdavila7/claude-code-templates	Skill	33k	yesterday	MIT	A	43	Claude Code, Codex

velocity.report: Skill for Claude Code